LAST spring, the 25-year-old Nuclear Non-Proliferation Treaty (NPT) was extended indefinitely. The NPT conference also adopted a Declaration of Principles on nuclear nonproliferation, which contains a call for an international treaty by 1997 banning all nuclear testing. Until then, the nuclear-weapon states have to exercise ''utmost restraint.'' A Comprehensive Test Ban Treaty (CTBT) has traditionally been advocated mainly as a means against vertical proliferation - weapons modernization by ''officially recognized'' nuclear-weapon states. In the NPT it is incorporated as a quid pro quo for other states' commitment to renounce nuclear weapons. The CTBT deadline was the only clearly defined arms-control measure the nuclear-weapon states were willing to accept at the conference. Over the last several months, certain developments have spoiled the euphoria over the extension of the treaty. Immediately following the conference, the Chinese conducted another nuclear test. Soon after that, the newly elected French president, Jacques Chirac, announced that his country will resume testing as well. French and Chinese motivations are probably alike. Both countries have emphasized that they want to join a CTBT, but both want to reach a level of technical sophistication at which they can rely on computer simulations to check and quite possibly modernize their nuclear arsenals before they renounce explosive testing. They may also want to catch up in testing technology because they hope to engage in any future arms reductions negotiations on a more or less equal footing with the United States and Russia. France, China may agree to ban To most people, it is a cynical paradox that France and China, in order to ratify a comprehensive test ban, are now implementing crash testing programs. What eventually may matter more, however, is the impact of their policy on the CTBT negotiations that started in Geneva early this summer. Will the street-wise diplomats at the Conference of Disarmament focus on Chinese and French testing per se, or will they be persuaded if these countries reconfirm a commitment to abide by a CTBT after 1996? Until recently, it also looked as if the negotiations might be wrecked by an American drive to include a threshold for explosive tests in the treaty. The Pentagon, especially, was insisting on continuation of fairly strong explosive tests at least for some years, arguing that these were needed for safety and reliability checks on existing arsenals. Even very low-yield, so-called ''hydronuclear'' tests, however, would put an ax to the roots of a CTBT. They would damage the very concept of comprehensiveness and they could provide new categories of data on weapons production to beginners. President Clinton wisely decided not to accept the Pentagon's advice, and to promote instead a ban on explosive testing that is as comprehensive as humanly possible. Now it is to be hoped Mr. Clinton will be able to get the Russians and the Chinese on his side. The French have already indicated their agreement (they probably feel they'll know all they need after finishing their current testing program), while the British don't have a choice, since they have always used US testing grounds for their explosions. Deadline is a key symbol As a practical instrument against nuclear proliferation, the value of a comprehensive test ban is limited at best. Advanced nuclear-weapon states can at least maintain their arsenals by computer simulations. Horizontal proliferation is not effectively impeded by a comprehensive ban either. Countries like Israel and Pakistan have created convincing nuclear options without explosive testing programs of their own. Over the years, however, establishing a CTBT has become immensely important as a political symbol. In this respect, failure of the Geneva negotiations would be ruinous. President Clinton, by his policy statement, has removed a major impediment. But insistence by some nuclear-weapon states on a testing threshold, in combination with the ongoing Chinese and French tests, may still prevent the conclusion of a CTBT in 1996. Then the only arms-control deadline agreed upon during the NPT extension conference would have been missed. And that in turn may have the even more serious effect of sapping the international nonproliferation regime of its persuasiveness. Much still depends on far-sighted American leadership.